Latest / Architecting the AI Enterprise / Episode 6: Trust but Verify, Reining In Overtrust in AI Outputs

Transcript

Episode 6: Trust but Verify, Reining In Overtrust in AI Outputs · Architecting the AI Enterprise

• Unqork: Welcome. I'm Gary Hoberman, CEO and founder of Uncorc. Hi, Dave Ferrucci, Chief Technology Officer and Chief AI Officer at Uncork. Yeah, we can't forget that, Dave. We gotta make sure we gotta keep adding to it. And ⁓ in this series, you know, architecting the AI Enterprise, we bring on top technology leaders, talk about what's real, what's happening, what's what's the future look like in their mind. Of course, all the disclaimers will put that we can't predict the future, but We do our best to try. And with that today, I've got a friend, colleague, partner. Like this is someone who I've known a long time is Steve Lord, global CIO of Access Capital. ⁓ welcome. Thank you. Thank you. I'm glad to be here. So, so Steve, like I knew Access Capital when the CIO was on the board with us back when I was a MetLife CIO. ⁓ you have to give the audience like a little bit of snapshot. Like, what is Access? Your new CEO is Amazing. Like I I look at CEOs and I look at like the caliber and out there, and he's definitely up there as far as vision goes. Please share a little bit more in Yeah, so we're we are a ⁓ a commercial insurer. We so we ⁓ specialty risks, these are complex commercial risks, and we write both insurance as well as reinsurance, where we're providing insurance to other insurance carriers. And we do this both in the US and Canada and North America as well as internationally. ⁓ so we have an entire global markets unit. So our risks run the gamut of of things that would be more interesting than your average. Can you give us examples of like what would be one craziest risk that you're Yeah, like ⁓ like when the ⁓ ship hit the Baltimore bridge and it collapsed, that is our loss. When ⁓ when somebody put put a ⁓ a missile into a ⁓ ship ⁓ docking at the United Arab Emirates in the Strait of Four Moose, that's us. So these are And we do we do directors and officers for public corporation. I mean, we do it's really a very wide gamut. Cyber cyber insurance. Cyber, of course, all forms of marine, the actual ships, their cargo, the liability, you name it. So this is this is this is not ⁓ mayhem or flow from progressive. I I can't imagine being the underwriter that looked at that and d you know, ⁓ yeah, that ship's not how do you decide if it's gonna hit There's a reason our stock goes up and down lately. ⁓ and ⁓ yeah it has to do with ⁓ whatever's ⁓ coming across the news that day. That's pretty amazing. And just give us a little background how you got to the global CIO position. 'Cause I I think we first met at Hartford, I think it was, but please let me No actually you know, we actually met my first go round at Axis. when I was working I I have had multiple roles in Axis and I'm a returning employee, yeah. So you and I met ⁓ I think a decade ago. And ⁓ at that time I was running the reinsurance IT unit and I was the CTO of the company. And then I I did leave and I ended up ⁓ at a company called Navigators, who's very similar to Axis in that it's a specialty insurer and reinsurer. Yes. Which was acquired by the Hartford. And then I had multiple roles at the Hartford and then returned to Axis when they were when they had a a CIO transition. And Vince when Vince Tizio came in as CEO, it kind of was a was a change for the company strategically. And I had worked with Vince ⁓ both at Navigators and at the Hartford. And so I was thrilled to come back and really have a chance to to ⁓ witness the company's almost rebirth, rebirth ⁓ under the new leader. I've gotten to meet Vince a few times. He has a vision and I love that there's no title like seniority. Like you don't feel he feels like he's just one of the guys you're talking to and it's just very easy. ⁓ yes, he's ⁓ he's very down to earth, very direct and ⁓ he is very driven. Yes. And so he's not he's not in this ⁓ for any reason other than to succeed. That's awesome. And by the way, before even then when we first met back at Access Capital version one, which is amazing, technology background? What was your what was so CTO, you had to have technical background for that? Yeah, I mean I me personally I've I started out in consulting, ⁓ doing systems integration consulting, writing code ⁓ for many years and I was e I'd even worked for a for a college professor who shipped ⁓ a logistics ⁓ product ⁓ bef in the years before Google Maps, truckers needed software to plot routes across the country. So that's where I first cut my teeth di kind of doing ⁓ enterprise software development for him and was in consulting for more than a decade. Wow, that's ⁓ definitely optimization challenge. Yeah, that's ⁓ it was. So so getting right into it. Like so AI has definitely changed the world. Like we we see it, it's exciting day in and day out. And every day there's some new announcement you didn't you couldn't predict. It's kinda like the boat hitting the the verge in some ways. ⁓ what are your views right now? What do you see happening within Access and what's I'm curious to see like what's what's active? And it's funny when when I When you and I were talking about AI a certain number of months ago and it was a certain number months ago, ⁓ you you've you were very, very ⁓ enthusiastic about its hate possibilities, and I was too, but lately I'm more I'm I'm catching up to you. I don't know that I'll ever quite be as enthusiastic as as as you are of a different person, but I I am very, very ⁓ excited about what its capabilities are. They've really turned a corner of From what I can see just in the last twelve to eighteen months that have just astonished even me. So I I mean, I think it's a real amazing opportunity to boost productivity both inside of inside of technology organizations like the one I lead, but also around it, ⁓ with our colleagues and partners. Yeah. I mean I ⁓ Dave likes I I guess Dave, your expression is it's ⁓ ⁓ we both what is it you say you both underestimate and overestimate. We we both tend to over and underestimate it at the same time. And you know, I mean I think sometimes we we overestimate it when we look at it as an oracle with the absolute correct answer to anything. And I think we underestimate it to your point, Steve, about how much of a productivity gain you can realize if you kinda know how to use it. know how to interact with it, help you formulate hypotheses, help you research, answer questions and formulate ideas and work products. But you know, you're really engaged with it. It's very different than going over, giving it a prompt and saying and then handing that to your boss or something like that. And so we underestimate in how much we can improve ⁓ not just the speed at which we do things, but the quality in which we do things if we in fact you look at it as like a thought partner as opposed to, you know, a crystal ball or something. It's kind of funny. This this week I was playing with the latest AI that we've been working with here. And, you know, Rich is in the room with us here, who's our chief evangelist head of sales, and I posted something in group going, you know, I think something happened because it's suddenly talking to me, the the AI is talking to me in c in what I thought was Chinese, Mandarin. And it just started doing it and I'm like speak Mandarin and I don't know it and Rich was like, actually it's Korean and it's cursing you out. So so Dave's point was, could you imagine giving access to that to do trades or doing something like that at that level? It's not ready yet, but ⁓ you know, it's interesting. So in the specialty market, like even the ⁓ insuring the cargo ships, like it could potentially look at ideas and help the underwriter think through God Here's all the things it knows about that area that maybe they didn't think of and I see I the the phrase I've been I've been using lately is we're trying to move from automation to augmentation. You know, for an underwriter, the benefits are playing the c the the ability and it's an it's kind of has a similar potential impact to anything where there's kind of a a a a a a niche or specialty skill set involved. Like it will it can augment you in a way where Hey, I I've compared this risk to all the other risks of this type that we've written. Here's the here's an in instant portfolio of how those have performed. Here are the the telltales, the giveaways that have that have been correlated this way or that way. So as an underwriter, you're starting out from the very first minute a turbocharge in a way that would have taken you a lot of personal effort and you have to be on your game to make sure that you do all of that. It's all possible to do. In certain cases just so difficult. That it's not feasible to do. Yes. It's not feasible to do. What's your you know, when you first re-engage with us when you came back to Access you brought your chief architect. Mm-hmm. Like that's what's your role of architecture? How do you view his role in the I so I've I've personally ⁓ acted in many of the roles of a technology professional. And so arc to me architecture is absolutely critical, but often overlooked by by by by the by various players in the ecosystem, people who are just wanting to build things, people who are just concerned about cost. Whereas I'm like, if you're gonna invest a a significant amount of money, say building a house, no one would ever think to to build a house without an architect. That will just we know it involves floors and walls and and plumbing will just get started and we'll just keep altering it as we go. We you know what kind of a house you'd end up with. So I'm ⁓ a very, very strong proponent of the I don't know why it pictured Dr. Seuss the houses in Dr. Seuss. It would end up like that. It would look like a shantytown type of thing. Lorox. ⁓ good so to me architecture, when you're mak investing the kind of ⁓ funds that you're investing typically in enterprise IT for a growing enterprise, you want to make sure that that to the extent possible you have both fle you you want to cover all the ⁓ the isms or the the itties that architecture does scalability sus ⁓ extensibility, ⁓ performance and so forth. You know, it's funny, the ilities is something which ⁓ most people don't understand unless you're leading technology. You're managing it. Like all the ilities. Yeah, no, that's that's it is critical. I mean I remember when it was created as a function. I ⁓ although recently I had the experience of it's smaller than a house, but I wanted to build a barbecue fire pit and like fireplace and pizza oven and Albrick and I took a picture of the area and I told ChatGPT here's what I want and it didn't and I said make it make it architectural plans and drawings. It did. I gave it to the Masonary and it was about two days later he came back to me and goes, I don't think that's accurate So I'm like Like, what do you mean? It looks cute. We're sitting such early days with the capabilities of these tools, yeah. For people of a certain age that remember say the the late nineties, early two thousands. I remember being in consulting when ⁓ the three G was coming around and a lot of folks were like, three G is really it's revolutionary and other countri when we still had flip phones, other countries they're already paying for per vending machines with phones, everything's gonna be on the phone. And yes, we had wireless, we had mobile, we had the internet, but no one was imagining that something like DoorDash was you know, an entire economy was coming From some of those core capabilities, which took a certain number of years later. And to me a lot of this still feels like it's it's very early day. It's still incredible and impressive, but being able to really predict all of the different permutations is a a tough job. If you were looking at it in terms of education, K through twelve, what grade are we in, do you think? ⁓ in the in the in the in the lifespan of of AI, we're first or second grade at the most, third grade. Dave, what do you I'm curious what you think. I've never asked you this. It's a good question. It's a I mean the journey I I mean it's ⁓ it it's an interesting question. I mean I I I there's two dimensions to that pop into my mind, which is the the the trajectory we're on with regard to how to use AI and they're related, and the trajectory for AI itself. Okay, yeah. And AI itself is moving quickly, but it's definitely relative to the last few years slowing down. I think we're going to be we're gonna see continued, you know, to lack of a better word, continued intelligence and capability. But I think we are further along on that trajectory than we are in figuring out how to even use it ⁓ effectively and efficiently. I think we're really Just starting. So if you because if you take the capability we have today, we're still grappling with what it is, how to integrate it into our business processes, and into our our enterprise systems, even into our personal lives, quite quite honestly. and I think we're really just scratching the surface on that. We're scratching the surface of what's possible. We are adopting the AI and figuring that is the bottleneck. And and I think there's also going to be ⁓ a ⁓ a positive feedback effect where once AI starts playing a bigger and more appropriate role, that itself is going to change the entire business process, enterprise operation ⁓ operations and implementation and there's gonna be a whole nother loop ⁓ that we're gonna go through. Don't you think Dave also like the way these these models right now and the the data centers needed to to build them and run them. Sometimes I think about like the early computers that were the size of a room. ⁓ cost a fortune and yeah. The form factor is NVIDIA shipping laptop chips for this it just it's interesting you say that because I I yeah, I go around, I give general, you know, more general talks to lay audiences on AI all the time. I love doing that 'cause I love shocking them. And now I bring up and I talk to, you know, GPT or whatever live. And it's so interesting that the response I get, you know, from the you know, from the audiences just in terms of how I use it is shocking to them. Cause I think the common way is to almost use it like a search system. And when they see me arguing with it, debating it, telling it its bias, and to remove its bias, and to use these sources and not those sources, and they look at me commanding it in a completely different way, and I get comments like you're changing my life. But one but one of the things that I open up with, and to your point, was I compare the ENIAC 1946 with the iPhone 16. The ENIAC ⁓ was five hundred floating point operations per second. The iPhone sixteen is thirty five trillion. Mm-hmm. Some insanity like that. And it and it fits in your hand, you can cake take it with you and everyone's got their own ⁓ E S in your correct. I go through all the stats, but I I end with this stat. A computation that would take a second today would take twenty five hundred years to compute. On the ENIAC eighty years ago. Yeah, nineteen forty six. Right. And so back in nineteen forty six with ENIAC, was it even I don't even know if science fiction ⁓ of that era really contemplated w what would be when the form factor and the cost per per calculation drop by so many orders of magnitude. So I think you're gonna see something ⁓ you're gonna see that trajectory. I don't know exactly what curve it'll follow, but I believe the cost will come down. Not not just because we're gonna continue to optimize the algorithms, but we're gonna compile and compress the data. There's gonna be all different ways in which to create and generate smaller models. I think the in the actual hardware for the inference systems at least and probably the training systems will get more and more efficient. ⁓ so you're gonna you're gonna see that curve. I think there's each interesting economics around it though that might slow that progress down because there's so much over investment right up front, right? I mean buying in other words, there was a massive investment in the scale and efficiency of a point in time. Yeah. You know what I think about a lot, Dave, is the again, I go back to what I know. If when the internet era started, there was tons of dark fiber, nobody could imagine how would we generate enough traffic to fill all this fiber. We figured it out. And then also when laptops became big, Intel pivoted their whole strategy to be power efficient. Yes. And actually switched their own architectures to chips that were insanely power efficient. When it was a desktop nobody cared. It's as long as it didn't melt how much power it used. They flipped their whole they flipped to their and now all the chips we use today are designed to be power efficient. And we haven't even We we we've barely even begun discussing those kinds of shifts for AI. Exactly. It'll take a certain number of years minimum, I imagine. So like and what will that mean when when those cur those those Moore's Law like curves start to come into play? We don't know. Correct. We don't know. I just know there'll be a lot more of it. Yeah, I mean and and and I think AI ⁓ what I heard was when AI takes over the physical shape of robots. They're gonna come after Dave before me and you because he's been arguing with I can't wait for the robot that can do laundry. Okay. I I I don't know I don't know. They may think the rest of you are a bunch of saps and they might have hired me. I don't I don't know. I'm happy to I'm happy to be a slave in that army dude. I I like it. It's good. I ⁓ it is it is a crazy world to predict. It's gonna be impossible like okay, IT function. Yeah. Five years. Where do you think? What what do you think the function's gonna be? I see, and I was talking about this with some of the team. I think there's always been the kind of ⁓ there's been role specialization. I'm an analyst, I'm a I'm a developer, I'm an architect, ⁓ I'm a manager, project manager in certain in certain modalities. But there's also some people who can do multiples of those things at once and with my sometimes you call them the ten X person or the rock star or whatever, and I think you're gonna see a lot more of the of the ⁓ l less specialization and more need for all of those skills in the folks that will be working with these types of AI powered augmentation alongside them ⁓ to generate good out outputs and outcomes bec at least at least for the next five years. I don't know after t they always say you c overestimate what'll happen in two and underestimate in ten. Five's right in the middle of that. So that's certainly moving faster than I've ever seen. You know, we heard it recently from a bank CIO ⁓ and they said To only hiring senior engineers now because they need engineers that could review the code being built in Telling it what to build, right? Like so you have it's not so easy to underwrite an ocean going vessel and understand the complexities of the business process. And if you just walk in off the street, then you might know ⁓ perfectly how to guide AI and code generation, but you still have to know what you're building and for what and And it's gonna be a long time before all of the existing sys I mean, I was a systems integration engineer for y ten years or more and I the work wasn't connecting systems together in ways the interface is not the issue, is making sure that they all work together smoothly day in and day out. And this these are whole new levels of problems beyond generating good architecturally sound code, which now becomes almost an afternoon. Yeah, in fact in fact, you know, the I I was recently reading some you know, estimates on this stuff, and again, take it with a grain of salt because no one really knows. But, you know, they were saying that that ratio of how much of a benefit you get with, for example, AI helping you write code, and if you break it up into simple applications, meeting and you know, super complex with thousands of function points and integrations and so forth and so on. and obviously as a result, larger human teams that It's a simple app, you know, you're cutting you're you're you're cutting it in half. ⁓ it r you know, you're sitting there with one or two people, the AI assistants, they're moving really, really quickly, you're getting a huge lift. As you go to medium size, that might turn out to twenty, twenty-five percent. And really complex apps is ten to fifteen percent. ⁓ because most of the hard work, as you're pointing out, Is not just in writing the code, it's figuring out what what do we even want it to do. Right, to find the problem. How do we sync up with a million other people? What are the integration points? What are the risk? It's all this other stuff that you have to get in sync on. And so the actual coding is just a less of a percentage of the overall effort. Yeah. I what I often say, what is the business problem we're trying to solve? And that comes from decades ago, but it's the same issue is what does the process we're trying to build look like? And I will say another thing is that while I think development will get faster and cheaper as it already is, yes. There's still the people around it that there's still there's still a cultural shift going on. You know, we're still you were saying, Dave, that people are amazed at watching you guide the AI and learning. Like those are the people that are even ⁓ fast followers that are are all over this, you know, this this stuff, ⁓ societally at least, when when grandma is interacting with it without even realizing it, you know, then we've kind of crossed the thresholds and we're not there. We're far from that. We're far from it. I ⁓ still chat thought mode these days. I mean Siri Siri's taken on a lot new since the past week, like I've seen the beta update. My sons my sons are all in AI, so I watched them play. They're interacting with Siri now the way I never thought they would be able to. And it has access to all of their data on their device, which is fascinating to think they're not thinking twice about take it all. Yeah. We all we're the old school. They used to say digital natives, now they're AI natives. AI natives. But there's still all the rest of us here. There's still ⁓ Can I ask on in a process inside of Access, is there a place today you could say you're comfortable putting an agent here to act? Yeah. Independent without Human in the loop without review, is there any place today or internal things, you know, internal things to it where the the agent ⁓ barrier is the actions that it can take and making its own decisions and so in in in insurance in general. You're so underwriters let's just pick underwriters, even claims professionals are conservative by nature. If you underwrite the wrong risk yes and that ship sinks, you just took, you know, a billion dollar loss type of thing. You know, that's So there's an inherent conservativism. And then we're a regulated industry. So you're not gonna let an agent there's always gonna be human in the loop for something that is transactional, financial, underwriting decisions of ⁓ of significance, claims handling. ⁓ I'd see a lot of agents that are more kind of informational, inserting agents into the, for example, ⁓ we were just talking about a wonderful idea to use agents to judge architectural. ⁓ compliance for code that's moving its way through the DevOps pipeline. And that's something all new. That's was just mind blowing to me. It literally was score the code. It was never possible it was never theor it was never something to think about because it just could not happen. But now it kind of can. There there are there are a lot of things like that that yeah, th those are fascinating. It's just we wouldn't even try to approach them, you know, with with traditional s ⁓ systems. Traditional software systems, yeah, it's quite quite interesting. I would never let an agent promote code to production on its own, you know. So anything that's gonna have the a significant downside, just like writing the wrong policy or the wrong limit, you're gonna want an H I T L situation there. Governance. You know, it has I mean, I get it, it has to be there, but it definitely is not the same thing as having the human do the original work. No. And it's just 'cause of it's just human cognitive orientation, right? I mean, there's a big difference between me checking someone else's work versus me doing it. Absolutely. And the more I trust that other person, the less work I actually do. So there's this risk of learning to generally trust the AI and missing those sorts of errors that have sort of asymmetric risks on your business. you know, it might be very easy to detect it if you paid attention. and so there really isn't all that much effort, but you end up not doing that effort 'cause you trust it, and then it's just asymmetric 'cause it's You mean like the people who are reading books behind the wheel, they're self driving cars. Yeah, I mean, you know, and then some instead of watching the road. And then something happens, right. So I th I think, you know, and again, ⁓ we don't know. We don't know how this is gonna play out and we need some empirical data. But it you know, there's it's when you yeah, as as I think we all know, like when you do some work yourself, you have a very different, intimate relationship with what what went on there. Yeah, I sometimes ⁓ I I never really paid any attention to human computer interaction ⁓ as a discipline. But I never think if it's about to get very, very interesting. It's interesting because we offshored development years and years ago. And we said, hey, go do this piece, give it back to us. We d let check it in, do code like in reality, we're trusting it was done well when it's a human. You could see it getting there, you know, but I to your point, even if it was creating the code, it still has to be checked before being released. Think about that's a good analogy, the offshoring analogy. Like the the the proponents of offshoring saw it as a panacea. Yes. Like we're just gonna send everything and it's gonna be great at whatever fraction of the cost. It didn't really work out. It it definitely was an augmentation, it was a an arbitrage. But show me the place that really has completely divorced have you ever seen their operator. And the captive was our employees. And they were I would say those employees were as strong as my team. When the market fell In the financial crisis, we sold the captive to a very large, you know, offshore firm. And we started to see turnover of forty, fifty percent of our employees as they were now part of that consulting company. And I'll never forget my senior engineers coming to me and going, Three to one. I'm like, What do you mean three to one? They're like, I'll give you back three of those headcount for one junior person here. And I did it. I said financially it makes it's a wash. It found a new equilibrium, but it was not like a just a set fire and forget solution. What was your experience? Have you seen it working and I I've only ever heard of the legends of the, you know, the ninety percent outsource thing, but t I only believe that ⁓ that's di f possible under laboratory conditions that don't exist in a growing company with complex and changing needs, which is all I I've ever wanted to work in. Like if you're working in a runoff insurer, you have different IT needs than it one that's adding products that is overhauling its ⁓ growing its practices, changing, and there are new forms of of business that are happening all the time. We're seeing kind of shifts into delegated underwriting. So there's constantly this is why people love talking to me at cocktail parties, because I I'm such a big insurance buff, but there's a lot of things going on. Yeah. And that that translates into the need for kind of high food chain intelligence in software development and support too. for that matter. I've never met anyone, Steve, besides you, who will say that they're the cocktail hour, like they're the ⁓ they're the cocktail hour. All I have to say at the cocktail party is I work at an IT for insurance and somebody needs to get to the bar and refill their drink immediately. There was actually a friend of mine, she's a she's an author in insurance. She's she was ⁓ she'd be brought in to do innovation in companies when I was at MetLife. She was but she used to fly all the time and I remember she gave me a cover of a book. It was just a cover and she sold them. And the cover of the book was called like the five most interesting things about life insurance. And she would put it around her books that she was actually interested in, just so people wouldn't bother her. You know. ⁓ I see. So it's sort of like an like a like an amulet to ward off people. It was a it was exact she said when I fly I wanna be by myself and that book covered I will do it. That's one person you don't want to start a random conversation. Maybe we should sell these. This could be a good thing. Life insurance. ⁓ But I mean specialty insurance is a lot more interesting than life insurance. Yes, it is. But anyway, obviously those those anecdotes which are real, they're like I said, they're I do it with my daughters, ⁓ although I say what's watch television, like every other commercial is a an insurance commercial. We have they have that most incredible marketing on the personal line stuff because they have to differentiate. But commercial line is a little different. Most most most average people aren't buying ⁓ you know, a side A director's and officers policy because you're on a board of a Fortune fifty. Correct types of thing. Correct. That's ⁓ wow that's okay. So let's go through. I think is there any other questions, Dave, on your side? Otherwise we'll go into lightning round if we're we're almost ready for that. Let's see. You wanna you wanna do the first lightning? You're gonna go first this time. I always go first. Come on. I I I always love to ask, you know, what you've binge w binge watched recently or what you or what you think is really good. you know, there's a couple of things that ⁓ that I've been burning through. One is ⁓ Your friends and neighbors. Yep. ⁓ which is a great show. So this kind of show is kind of like ⁓ almost a cinema almost verite of life in lower southern Westchester County where I live. so you know what I'm moving in near you. I'm coming away in a few weeks. Okay. So it's this is a little bit exaggerated, but it's filmed in the area and so it's it's riveted. I like the actors, so I enjoy I ⁓ enjoy that show. And then I've been watching ⁓ I've been I just picked up Widows Bay. Which is ⁓ which is a comedy horror ⁓ or horror comedy, so ⁓ so to speak, and it's ⁓ it's very subtle and very cute. I I happen to think Friends and Neighbours is one of the best shows. It is, it's I actually just I though I'm referencing it like Seinfeld lately. Yeah. Like again, I don't think Seinfeld has ever had a show a c like it. And for the younger generation, they're like what Seinfeld? But I use it in all analogies. We always do. ⁓ give me your favorite book that you'd recommend. So ⁓ one of the ones A book that I personally enjoy. I I do I as the longer I I ⁓ work professionally, the more I pick up kind of management management tomes and management books and twenty years ago everyone was talking about good to great. It's kind of these kind of arch and one of the more recent ones is by an author named Liz Wiseman called Multipliers. I don't know if you've heard of it or not, but ⁓ the concept is around leaders that improve their organizations by mul they're multiplying the capabilities and then their inverse, their doppels doppelgangers are diminishers, those that by their actions and by their characteristics diminish the the the capabilities of those that they work with. And I thought this was just a something when you read it and you go, yes, this is I I know people, I see these these things and now I have a a f mental framework that neatly encapsulates it. And it's become a shorthand for me. I love behavioral economics books. I love the idea. This is that's what that that's and Dave. Yeah. Think fast, think slow. And then I also and I love Daryl Ariella, right? ⁓ yes. Very good too, yeah. Yeah. I mean I I think it's fascinating how crazy we actually are. So well are we know I I work in IT, so I was never the soft skills were always the the lesser priority w I always tell people I went into tech not because I was good at dealing with people. And yet at this ⁓ you know, in today's world of enterprise IT, dealing with people is the job. It's the job. It's all about it's getting Dave, I'm gonna ask you one which we didn't which is are we li what's the probability we're living in a simulation with that point? When I said people are crazy and we're talking AI, ⁓ give it a number. ⁓ I don't know. I think it's a very I I think it's a It's ⁓ it's a fascinating theory and it's very h it's very hard to argue that it's it's very hard to argue that it's not plausible. I just don't know how to give it I I'm not sure how to give it a probability. Isn't that kinda like Pascal's wager? If you believe in it strongly, what's the point of anything? You know, like yeah it's kind of the inverse of Pascal's. I keep picturing there's some kid with Pimples that's controlling me behind the scenes like at a game playing sim. Like I it's it's interesting. It is a very it's very it's a very strong, strong analogy. It's very interesting. ⁓ it holds up under so it holds up surprisingly under late deep scrutiny anyway. It's who knows? That's what they they say eventually with AI we might find the answer to that. We'll have to see if we're ⁓ but Steve, this was amazing. Again, like ⁓ your leadership and getting to watch and you know, your whole journey is incredible and ⁓ appreciate you coming here and now our office. You didn't push a button, you could push a button before we go, 'cause we got this new control panel. So let's just do it. Yeah. Awesome. That seems appropriate at this point in the in the discussion. Awesome. Steve, thank you for for being an amazing friend partner and ⁓ and being here with us. ⁓ appreciate the opportunity and ⁓ and I'm sh I will be back for sure. Awesome. Again. And thank you, ⁓ thanks everyone for watching. Thanks, Dave. And until next time, ⁓ Architecting AI Enterprise. Take care everyone thank you. Thank you to our listeners for tuning in. Make sure to like and subscribe. We hope to see you again in our next episode. Please send ideas for topics or guests that you'd like to hear from. We look forward to seeing you then. Thanks.